Quail: open-source AI-SQL engine hits 1B+ tokens/min on a single H100
sh_reya · x · 2026-09-25
shreya and Modal released Quail, an open-source AI-SQL engine built by co-designing query planning with LLM inference. AI-powered data operators form a novel inference workload, and by planning queries and inference together Quail reaches over 1 billion input tokens per minute on a single H100 for a single query — a showcase of how far inference-stack/use-case co-design can push throughput.
More from coding & agent
- Jev Founder's Agent Harness Guide Claims 200x Faster, 400x Cheaper Coding Agents — blaizedsouza · 2026-09-25
- Foreman: An Open-Source Jev-Powered Supervisor Loop Wrapped Around Codex — blaizedsouza · 2026-09-25
- Kagriculture agent contest becomes largest competition in Kaggle history — MeganRisdal · 2026-09-25
- Jev Founder's 12-Page PDF: 10-Step Blueprint for a Fast Decision Layer Around Any LLM — blaizedsouza · 2026-09-25
- Jev as Judge: Structured Decisions Replace Text Generation for Agent Evals, With Open-Source Demo — blaizedsouza · 2026-09-25
- Developers hunt for the gnarliest 'unmergeable' AI-generated code slop screenshots — pvncher · 2026-09-25